Arbitrage Watch

Advanced statistics and complex mathematics doesn’t help and shift your focus and energy.

You should not worry so much about backtesting. Backtesting is a big time waster and it will not help your trading. Backtesting is curve fitting which means that you are basically testing all these parameter values based upon noisy data. You should instead figure out the theoretical reason why an specific strategy should be profitable and then you should focus your energy in finding such trades. Backtesting, advanced statistics, complex mathematics will hypnotize you and shift your focus from trading to BS.

Once you have a theory, backtesting is an essential step in checking it. Of course, this doesn’t guarantee anything but if, for example, your strategy has not worked in the past, you at least have to explain what’s “different this time”: it’s very useful to “kick the tires” in this way. Ironically, it’s almost more dangerous if your backtest results are very good: that tends to give people false confidence. That being said, if your strategy has worked in backtests, it does give you some support to your theory.

Furthermore, backtesting is one of the ways to “avoid being hypnotized” by theory precisely because it is so concrete. There is no way - with any theory - to avoid the past-doesn’t-predict-the-future/in-vs-out-of-sample problem but, if your theory is to stand a chance it must, at least, be consistent with what has happened so far.

There are contexts in which you need to be more careful - particularly, if your trading is likely to change trading in the rest of the market - but, certainly, if your strategy going to be low-key enough not to move prices too much or have other strategic effects, backtest results are always useful information.